In this series of posts:
Carl B. Frey & Michael A. Osborne. The Future of Employment: How susceptible are jobs to computerization? Technological forecasting and social change. 2017. Published version here. Open access working paper here.
Melania Arntz, Terry Gregory & Ulrich Zierahn. The risk of automation for jobs in OECD countries. OECD working paper available here.
David H. Autor. Why are there still so many jobs? The history and future of workplace automation. Journal of economic perspectives. 2015. Published, open access version available here.
Susan F. Lu, Huaxia Rui & Abraham Seidman. Does technology substitute for nurses? Staffing decisions in nursing homes. Management Science. Forthcoming. Published open access version available here.
Reading the projections about jobs and technology left some sense of anxiety that technology will replace at least a good share of jobs – how it will play out is up for grabs, but substitution is certainly coming.
David Autor in his imminently readable piece (and talk) points out that labor-saving substitution (automating some work, e.g. calculation) and technological job destruction (making some work obsolete, e.g. shoeing horses) has been going on since at least the Industrial Revolution. And yet we still have jobs. Perhaps more jobs than we need and too much obsession about productivity in those jobs, but jobs nonetheless.
Why is that? Because of complementarity and demand.
As he puts it “tasks that cannot be substituted bby automation are generally complemented by by” (p. 6). In other words, if some aspect of your job can be automated and become more productive, the value of those aspects that can’t be automated might increase (i.e. be complemented), because their relative importance to overall productivity increases. The applies, however, only if your primary tasks or the tasks that you are best at are not substitutable. This reflects what’s called ‘the O-Ring’ production function (formally described, less formally described), suggesting that a seemingly unimportant component can be highly important if everything else works very smoothly.
Here’s an example: ATMs replace the work done by bank tellers. And yet there are more bank tellers working in US banks today than before ATMs. One of the reasons for that is that bank tellers now do a different job, where they spend less time on mundane things like helping people deal with cash and spend more time on advising customers. The relational part of bank tellers’ work has become more valuable. This means that the skills required to be a bank teller change (many bank tellers now have a college education). Similar (upskilling) changes in the content of work seem to also occur in other fields, even if it does not always cause increased demand for workers.
At the same time, ATMs let you operate a bank branch with fewer employees, meaning that banks find it profitable to open up more branches and thus need more bank tellers. This relates to the question of demand. In the bank teller case, customers seem to want more relational work from their bank tellers. More generally, if it becomes possible to produce something more efficiently and hence more cheaply, people may demand more of it. To use a previous reference, if silk stockings can be produced more cheaply, they will no longer be consumed only by royalty. If some aspect of silk stocking production has to be done by hand, it is clear that there will be more demand for workers to do that aspect of production. Moreover, increased efficiency in one area of the economy allows people to spend the money they save on other things, thus boosting demand. Demand, it turns out, is tremendously ‘elastic‘ and so when people earn more, they also consume more. They do not just consume the same and work less, which I understand (technology is, after all, giving us more and more really great stuff to consume, even if one can validly question whether advertising is producing false needs and consumerism is bad).
So, in this sense, there will continue to be jobs. Autor is sensitive (sort of and not very sociologically) to the adjustment problem (“young US adults… have responded with remarkable sluggishness to the rising educational premium…” p. 16) and the issue of wage polarization, but fundamentally optimistic that there will be jobs, also in the future.
Autor is also very sensitive to context – particular features of individual jobs and markets really matter. Next post, we’ll take a look at one example of such a context, specifically how and when technology substitutes for nurses.